Deployable AI for Public Safety: Weapon Detection in Challenging CCTV Scenarios
Bibliographic record
Abstract
Weapon detection in CCTV surveillance is a critical task for enhancing public safety, where real-time accuracy and robustness directly influence prevention and response efforts. Existing deep learning solutions often demonstrate strong results on clean datasets but degrade significantly under real-world conditions involving low resolution, poor lighting, motion blur, or partial occlusion. This paper investigates the performance of the latest YOLOv12n model for detecting pistols and knives in CCTV imagery, benchmarking it against YOLOv8n under identical training conditions. A composite dataset of 8,065 annotated CCTV-style images from Kaggle and Roboflow was used, encompassing diverse lighting and crowded environments. Preprocessing steps include resizing, normalization, and augmentation to improve generalization. We evaluated all models by varying input resolutions, confidence thresholds, and training epochs. For assessing performance, we assessed it by using mAP@0.5, precision-recall curves, and F1 confidence analysis. Results show YOLOv12n outperforming YOLOv8n with a validation mAP@0.5 of 0.716 after 15 epochs, where it achieved strong knife detection but less reliable pistol recognition, especially in low-light scenarios. We noted that size of image had little effect on accuracy but it significantly impacts speed of inference. Smaller sized images helped in faster detection.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".